TASI Lectures on Solitons Lecture 3: Vortices
Preprint typeset in JHEP style - PAPER VERSION June 2005
TASI Lectures on Solitons Lecture 3: Vortices
David Tong Department of Applied Mathematics and Theoretical Physics, Centre for Mathematical Sciences, Wilberforce Road, Cambridge, CB3 OBA, UK [email protected]
Abstract: In this third lecture we turn our attention to vortex strings in non-abelian gauge theories. Once again we describe the various properties of the solutions and a D-brane construction of the vortex moduli space is reviewed. In this lecture we will see how instantons can be viewed from the worldsheet of a vortex string, an important result that underlies the relationship between 2d sigma-models and 4d gauge theories. We end with the usual vignettes, describing applications of vortices to cosmic strings and mirror symmetry. Contents
3. Vortices 2 3.1 The Basics 2 3.2 The Vortex Equations 4 3.3 The Moduli Space 6 3.3.1 The Moduli Space Metric 7 3.3.2 Examples of Vortex Moduli Spaces 8 3.4 Brane Construction 10 3.4.1 Examples of Vortex Moduli Spaces Revisited 13 3.4.2 The Relationship to Instantons 14 3.5 Adding Flavors 15 3.5.1 Non-Commutative Vortices 17 3.6 What Became Of...... 17 3.6.1 Monopoles 17 3.6.2 Instantons 18 3.7 Fermi Zero Modes 19 3.8 Applications 20 3.8.1 Vortices and Mirror Symmetry 21 3.8.2 Swapping Vortices and Electrons 21 3.8.3 Vortex Strings 23
–1– 3. Vortices
In this lecture, we’re going to discuss vortices. The motivation for studying vortices should be obvious: they are one of the most ubiquitous objects in physics. On table- tops, vortices appear as magnetic flux tubes in superconductors and fractionally charged quasi-excitations in quantum Hall fluids. In the sky, vortices in the guise of cosmic strings have been one one of the most enduring themes in cosmology research. With new gravitational wave detectors coming on line, there is hope that we may be able to see the distinctive signatures of these strings as the twist and whip. Finally, and more formally, vortices play a crucial role in determining the phases of low-dimensional quantum systems: from the phase-slip of superconducting wires, to the physics of strings propagating on Calabi-Yau manifolds, the vortex is key.
As we shall see in detail below, in four dimensional theories vortices are string like objects, carrying magnetic flux threaded through their core. They are the semi-classical cousins of the more elusive QCD flux tubes. In what follows we will primarily be inter- ested in the dynamics of infinitely long, parallel vortex strings and the long-wavelength modes they support. There are a number of reviews on the dynamics of vortices in four dimensions, mostly in the context of cosmic strings [1, 2, 3].
3.1 The Basics In order for our theory to support vortices, we must add a further field to our La- grangian. In fact we must make two deformations We increase the gauge group from SU(N) to U(N). We could have done this • before now, but as we have considered only fields in the adjoint representation the central U(1) would have simply decoupled.
We add matter in the fundamental representation of U(N). We’ll add N scalar • f fields qi, i =1 ...,Nf . The action that we’ll work with throughout this lecture is
N 1 1 f S = d4x Tr F µνF + ( φ)2 + q 2 2e2 µν e2 Dµ |Dµ i| Z i=1 N N X f e2 f q†φ2q Tr ( q q† v2 1 )2 (3.1) − i i − 4 i i − N i=1 i=1 X X The potential is of the type admitting a completion to =1or = 2 supersymmetry. N N In this context, the final term is called the D-term. Note that everything in the bracket
–2– of the D-term is an N N matrix. Note also that the couplings in front of the potential × are not arbitrary: they have been tuned to critical values.
We’ve included a new parameter, v2, in the potential. Obviously this will induce a vev for q. In the context of supersymmetric gauge theories, this parameter is known as a Fayet-Iliopoulos term.
We are interested in ground states of the theory with vanishing potential. For Nf < N, one can’t set the D-term to zero since the first term is, at most, rank Nf , while the v2 term is rank N. In the context of supersymmetric theories, this leads to spontaneous supersymmetry breaking. In what follows we’ll only consider N N. In fact, for the f ≥ first half of this section we’ll restrict ourselves to the simplest case:
Nf = N (3.2)
With this choice, we can view q as an N N matrix qa , where a is the color index and × i i the flavor index. Up to gauge transformations, there is a unique ground state of the theory,
a a φ =0 , q i = vδ i (3.3)
Studying small fluctuations around this vacuum, we find that all gauge fields and scalars are massive, and all have the same mass M 2 = e2v2. The fact that all masses are equal is a consequence of tuning the coefficients of the potential.
The theory has a U(N) SU(N) gauge and flavor symmetry. On the quark fields G × F q this acts as
q UqV † U U(N) , V SU(N) (3.4) → ∈ G ∈ F The vacuum expectation value (3.3) is preserved only for transformations of the form U = V , meaning that we have the pattern of spontaneous symmetry breaking
U(N) SU(N) SU(N)diag (3.5) G × F → This is known as the color-flavor-locked phase in the high-density QCD literature [4].
When N = 1, our theory is the well-studied abelian Higgs model, which has been known for many years to support vortex strings [5, 6]. These vortex strings also exist in the non-abelian theory and enjoy rather rich properties, as we shall now see. Let’s choose the strings to lie in the x3 direction. To support such objects, the scalar fields q 1 1 2 must wind around S∞ at spatial infinity in the (x , x ) plane, transverse to the string.
–3– As we’re used to by now, such winding is characterized by the homotopy group, this time
Π1 (U(N) SU(N)/SU(N)diag) = Z (3.6) × ∼ which means that we can expect vortex strings supported by a single winding number k Z. To see that this winding of the scalar is associated with magnetic flux, we use ∈ the same trick as for monopoles. Finiteness of the quark kinetic term requires that q 1/r2 as r . But a winding around S1 necessarily means that ∂q 1/r. To D ∼ →∞ ∞ ∼ cancel this, we must turn on A i∂q q−1 asymptotically. The winding of the scalar at → infinity is determined by an integer k, defined by
−1 1 2 2πk = Tr i∂θq q = Tr Aθ = Tr dx dx B3 (3.7) S1 S1 I ∞ I ∞ Z This time however, in contrast to the case of magnetic monopoles, there is no long range magnetic flux. Physically this is because the theory has a mass gap, ensuring any excitations die exponentially. The result, as we shall see, is that the magnetic flux is confined in the center of the vortex string.
x The Lagrangian of equation (3.1) is very spe- 3 cial, and far from the only theory admitting vor- tex solutions. Indeed, the vortex zoo is well pop- ulated with different objects, many exhibiting cu- rious properties. Particularly interesting examples include Alice strings [8, 9], and vortices in Chern- Simons theories [10]. In this lecture we shall stick phase of q with the vortices arising from (3.1) since, as we shall see, they are closely related to the instantons Figure 1: and monopoles described in the previous lectures. To my knowledge, the properties of non-abelian vortices in this model were studied only quite recently in [11] (a related model, sharing similar properties, appeared at the same time [12]).
3.2 The Vortex Equations To derive the vortex equations we once again perform the Bogomoln’yi completing the square trick (due, once again, to Bogomoln’yi [7]). We look for static strings in the x3 direction, so make the ansatz ∂0 = ∂3 = 0 and A0 = A3 = 0. We also set φ = 0. In fact φ will not play a role for the remainder of this lecture, although it will be resurrected
–4– in the following lecture. The tension (energy per unit length) of the string is
2 N N 1 2 1 2 e † 2 2 2 2 Tvortex = dx dx Tr B + ( q q v 1 ) + 1q + 2q e2 3 4 i i − N |D i| |D i| Z i=1 ! i=1 X 2 X 2 N N 1 2 1 e † 2 2 = dx dx Tr B3 ( q q v 1 ) + 1q i 2q e2 ∓ 2 i i − N |D i ∓ D i| i=1 ! i=1 Z X X 2 1 2 v dx dx Tr B3 (3.8) ∓ Z To get from the first line to the second, we need to use the fact that [D1,D2]= iB3, − to cancel the cross terms from the two squares. Using (3.7), we find that the tension of the charge k vortex is bounded by | | 2 Tvortex 2πv k (3.9) ≥ | | In what follows we focus on vortex solutions with winding k < 0. (These are mapped into k > 0 vortices by a parity transformation, so there is no loss of generality). The inequality is then saturated for configurations obeying the vortex equations
2 e † 2 B3 = ( q q v 1 ) , q =0 (3.10) 2 i i − N Dz i i X where we’ve introduced the complex coordinate z = x1 + ix2 on the plane transverse to 1 the vortex string, so ∂ = (∂1 i∂2). If we choose N = 1, then the Lagrangian (3.1) z 2 − reduces to the abelian-Higgs model and, until recently, attention mostly focussed on this abelian variety of the equations (3.10). However, as we shall see below, when the vortex equations are non-abelian, so each side of the first equation (3.10) is an N N × matrix, they have a much more interesting structure.
Unlike monopoles and instantons, no analytic solution to the vortex equations is known. This is true even for a single k = 1 vortex in the U(1) theory. There’s nothing sinister about this. It’s just that differential equations are hard and no one has decided to call the vortex solution a special function and give it a name! However, it’s not difficult to plot the solution numerically and the profile of the fields is sketched below. The energy density is localized within a core of the vortex of size L =1/ev, outside of which all fields return exponentially to their vacuum.
The simplest k = 1 vortex in the abelian N = 1 theory has just two collective coordinates, corresponding to its position on the z-plane. But what are the collective coordinates of a vortex in U(N)? We can use the same idea we saw in the instanton
–5– 2 |q| B3
v2
r r 1/ev 1/ev
Figure 2: A sketch of the vortex profile. lecture, and embed the abelian vortex — let’s denote it q⋆ and A⋆ — in the N N z × matrices of the non-abelian theory. We have
⋆ ⋆ Az q 0 v Az = . , q = . (3.11) .. .. 0 v where the columns of the q matrix carry the color charge, while the rows carry the flavor charge. We have chosen the embedding above to lie in the upper left-hand corner but this isn’t unique. We can rotate into other embeddings by acting with the SU(N)diag symmetry preserved in the vacuum. Dividing by the stabilizer, we find the internal moduli space of the single non-abelian vortex to be
N−1 SU(N)diag/S[U(N 1) U(1)] = CP (3.12) − × ∼ The appearance of CPN−1 as the internal space of the vortex is interesting: it tells us that the low-energy dynamics of a vortex string is the much studied quantum CPN−1 sigma model. We’ll see the significance of this in the following lecture. For now, let’s look more closely at the moduli of the vortices.
3.3 The Moduli Space We’ve seen that a single vortex has 2N collective coordinates: 2 translations, and 2(N 1) internal modes, dictating the orientation of the vortex in color and flavor − space. We denote the moduli space of charge k vortices in the U(N) gauge theory as . We’ve learnt above that Vk,N N−1 1 = C CP (3.13) V ,N ∼ ×
–6– What about higher k? An index theorem [14, 11] tells us that the number of collective coordinates is dim( )=2kN (3.14) Vk,N Look familiar? Remember the result for k instantons in U(N) that we found in lecture 1: dim( )=4kN. We’ll see more of this similarity between instantons and vortices Ik,N in the following.
As for previous solitons, the counting (3.14) has a natural interpretation: k parallel vortex strings may be placed at arbitrary positions, each carrying 2(N 1) independent − orientational modes. Thinking physically in terms of forces between vortices, this is a consequence of tuning the coefficient e2/4 in front of the D-term in (3.1) so that the mass of the gauge bosons equals the mass of the q scalars. If this coupling is turned up, the scalar mass increases and so mediates a force with shorter range than the gauge bosons, causing the vortices to repel. (Recall the general rule: spin 0 particles give rise to attractive forces; spin 1 repulsive). This is a type II non-abelian superconductor. If the coupling decreases, the mass of the scalar decreases and the vortices attract. This is a non-abelian type I superconductor. In the following, we keep with the critically coupled case (3.1) for which the first order equations (3.10) yield solutions with vortices at arbitrary position.
3.3.1 The Moduli Space Metric There is again a natural metric on arising from taking the overlap of zero modes. Vk,N These zero modes must solve the linearized vortex equations together with a suitable background gauge fixing condition. The linearized vortex equations read 2 ie † † δA¯ ¯δA = (δq q + q δq ) and δq = iδA q (3.15) Dz z −Dz z 4 Dz z where q is to be viewed as an N N matrix in these equations. The gauge fixing × condition is 2 ie † † δA¯ + ¯δA = (δq q q δq ) (3.16) Dz z Dz z − 4 − which combines with the first equation in (3.15) to give 2 ie † ¯δA = δq q (3.17) Dz z − 4
Then, from the index theorem, we know that there are 2kN zero modes (δαAz, δαq), α, β =1,..., 2kN solving these equations, providing a metric on defined by Vk,N 1 1 g = Tr dx1dx2 δ A δ A + δ qδ q† + h.c. (3.18) αβ e2 α a β z¯ 2 α β Z
–7– The metric has the following properties [15, 16] The metric is K¨ahler. This follows from similar arguments to those given for • hyperK¨ahlerity of the instanton moduli space, the complex structure now de- scending from that on the plane R2, together with the obvious complex structure on q.
The metric is smooth. It has no singularities as the vortices approach each other. • Strictly speaking this statement has been proven only for abelian vortices. For non-abelian vortices, we shall show this using branes in the following section.
The metric inherits a U(1) SU(N) holomorphic isometry from the rotational • × and internal symmetry of the Lagrangian.
The metric is unknown for k 2. The leading order, exponentially suppressed, • ≥ corrections to the flat metric were computed recently [17].
3.3.2 Examples of Vortex Moduli Spaces A Single U(N) Vortex We’ve already seen above that the moduli space for a single k = 1 vortex in U(N) is
N−1 1 = C CP (3.19) V ,N ∼ × where the isometry group SU(N) ensures that CPN−1 is endowed with the round, Fubini-Study metric. The only question remaining is the size, or K¨ahler class, of the CPN−1. This can be computed either from a D-brane construction [11] or, more conventionally, from the overlap of zero modes [18]. We’ll see the former in the following section. Here let’s sketch the latter. The orientational zero modes of the vortex take the form
δA = Ω , δq = i(Ωq qΩ0) (3.20) z Dz − where the gauge transformation asymptotes to Ω Ω0, and Ω0 is the flavor transfor- → mation. The gauge fixing condition requires 2 2 e † † Ω= Ω, qq 2qq Ω0 (3.21) D 2 { } − By explicitly computing the overlap of these zero modes, it can be shown that the size of the CPN−1 is 4π r = (3.22) e2 This important equation will play a crucial role in the correspondence between 2d sigma models and 4d gauge theories that we’ll meet in the following lecture.
–8– Two U(1) Vortices The moduli space of two vortices in a U(1) gauge theory is topologically
=2 =1 = C C/Z2 (3.23) Vk ,N ∼ × where the Z2 reflects the fact that the two solitons are indistinguishable. Note that the notation we used above actually describes more than the topology of the manifold k k because, topologically, C /Zk ∼= C (as any polynomial will tell you). So when I write C/Z2 in (3.23), I mean that asymptotically the space is endowed with the flat metric on C/Z2. Of course, this can’t be true closer to the origin since we know the vortex moduli space is complete. The cone must be smooth at the tip, as shown in figure 3. The metric on the cone has been computed numerically [19], no analytic form is known. The deviations from the flat, singular, metric on the cone are exponentially suppressed and parameterized by the size of the vortex L 1/ev. ∼